Psychotherapy logo
John Neumin Psychotherapy All Photos by John Neumin psychotherapy

Home Psychotherapy John Neumin Resources FAQ Contact

Is Psychotherapy Effective?

What Can Efficacy Studies Tell Us?


If you are looking to begin psychotherapy and investigating it on the web, you may be confused by the mass of information and opinion available to you.  You may have read some of the many contradictory, ambiguous and politicized claims about its effectiveness, ineffectiveness or relative effectiveness.  You may feel less sure about whether psychotherapy can help you than when you started. 

I believe that the controversy regarding the efficacy of psychotherapy is based on imprecise definitions, faulty logic, weak scientific thinking and, ultimately, a profound vagueness as to what is actually being asked, and I will try to show that a statistical approach to resolving this issue is not simply flawed, but will inevitably fail to provide meaningful results.

In 1952, Hans Eysenck changed the terms on which therapeutic efficacy would be debated. In analyzing 24 studies that seemingly indicated that there was no difference in outcomes between patients who received psychotherapy and those who did not.  His paper was, according to David G. Myers (1998), “the opening volley in a spirited debate”.  Eysenck not only challenged psychotherapy, but he did so statistically and since that time, authors on either side of the debate have used statistical analysis to support their findings, with neither side able to provide clear evidence for its position.  Study after study is brought forward allegedly confirming one position or the other, or touting the superior efficacy of one type of therapy over another. Meta-analyses (such as Smith & Glass, 1977) are developed in an attempt to make sense of the mass of contradictory findings, and criticisms and re-evaluations of the meta-analyses are brought forward to provide evidence to support the opposite position.  Ultimately little progress is made in arriving at a consensus.

In 1952, in direct response to Eysenck’s study, Harry Guntrip (1952), an influential British analyst, challenged many of the basic assumptions that were implicit in the use of statistical studies to try to investigate the efficacy of psychotherapy. Guntrip criticized statistical analysis on a number of points. He asked -- Are the patients that are ‘cured’ without psychotherapy really in the same state as those ‘cured’ through psychotherapy?  Or in other words, are all ‘cures’ equal.  What criteria do we use to decide whether someone is ‘cured’? And, how do we observe and measure what are very likely to be complex, subtle and often mostly subjective markers of improved mental health?  Prioleau, Murdock and Brody (1983), in their own statistical study -- one that they argue disputes the value of psychotherapy -- nevertheless acknowledge shortcomings of their method that closely parallel Guntrip’s criticism of statistical bluntness. They say:

The weakest aspect of outcome research involves the measures used to assess change. It is possible that psychotherapy produces beneficial changes … but the available outcome measures are too crude to detect such differences. Perhaps the major respect in which outcome research could be improved would be through the use of a wider range of dependent variables with more power to detect (possibly subtle) differences. (p. 284)

In agreement with the aforementioned authors, others (Yalom, 1980, Breggin, 1991) note that it is very difficult both to determine and to measure significant markers of effectiveness, and many (including Guntrip, Yalom and Breggin) believe that this problem is an impossible one to address.  Yalom, once a proponent of statistical research, explains his reevaluation of its effectiveness regarding psychotherapy:

I could not believe that the true experience of the participants was adequately described by our highly technical, computerized statistical approach . . . Again and again one encounters a basic fact of life in psychotherapy research: the precision of the result is directly proportional to the triviality of the variables studied. (p. 24)

            As strong as these arguments are however, Guntrip’s final criticism of Eysenyck’s use of statistics is an even more basic one. He says “The most serious omission in any statistical analysis of therapeutic results [is that] it fails to take into consideration the nature of the relationship between psychotherapist and patient, but treats psychotherapy as if it were a fixed and known entity, the same thing in every case” (p. 67). Guntrip questions the vagueness of the definition of ‘psychotherapy’ and states that ultimately psychotherapies cannot be compared, but are, as Freud noted fifty years before, “radically heterogeneous”.  He goes on to point out that over and above the fact that each therapist is different, is the fact that the patient and therapist need to be matched for best results, and he concludes that statistical analyses must fail because they can neither predict the many possible gains made by a client in psychotherapy, nor can they measure them, and most importantly, that ultimately it is invalid to compare one type of therapy to another, or even one therapeutic dyad to another.
How then can we answer the question “Is psychotherapy effective?” if not statistically? Do we discard it completely, and if not, how do we proceed? Ludwig Wittgenstein suggested that all seemingly unresolvable conceptual disagreements are indicators of inadequately precise language.

Returning to the original question “Is psychotherapy effective?” it must be concluded that those debating the two sides are locked in a fruitless debate.  This debate would be better served if all were to heed Wittgenstein’s warning.

 . . . the difficulty is not of finding the answer, but rather that of recognizing as the solution something that looks as if it were only a preliminary to it. . . the solution of the difficulty is a description, if we give it the right place in our considerations. If we dwell upon it, and do not try to get beyond it.  The difficulty here is: to stop. (p. 107)

The question as asked  - “Is psychotherapy effective?” - is vague, ambiguous, and even obfuscating, and like the questions “Is medicine effective?” “Is parenting effective?” or “Is life effective?” has no meaningful answer. Resolving the disagreements regarding whether psychotherapy works, or how well it works, requires finding finer definitions of what we are discussing, and more nuanced understandings of what actually takes place when two people enter a room to ‘do psychotherapy’. “The solution of the difficulty is a description.”   A description that allows us a broader, deeper and richer understanding of what psychotherapy is, what it isn’t, and what we need it to be.

Can a statistical study help you decide whether to begin therapy, what kind of therapy to try, and how to choose a therapist?  I believe the answer to these questions is no.  The nature of therapy is too varied, complex, and subjective; it cannot be measured by even the most sophisticated statistical analysis.  What you need in your therapy is unique.  What each therapist offers is just as unique.  I hope this short note has helped to clarify some of the possible confusions around the many claims you might read about efficacy, and I hope it has highlighted the importance of finding the right therapist for you - another varied, complex, subjective and nuanced process that no statistical analysis can illuminate.



"What lies behind us and what lies before us are tiny matters compared to what lies within us. "
Oliver Wendell Holmes

Copyright © John Neumin

website design by Turbo Computer Systems